Reused conv2d pattern matching checks

Change-Id: I1e032295668e68849a365da0dde28234c66cc056

python/tvm/relay/op/contrib/cmsisnn.py

@@ -164,16 +164,12 @@ def check_qnn_conv2d_pad(pattern):
         else:
             requantize = pattern
         requantize_input = requantize.args[0]
-        bias_add = None
-        bias_dtype = "int32"
         if str(requantize_input.op.name) == "nn.bias_add":
             bias_add = requantize_input
             conv2d = bias_add.args[0]
-            bias_dtype = bias_add.args[1].checked_type.dtype
         else:
             conv2d = requantize_input
         conv2d_input = conv2d.args[0]
-        conv2d_weight = conv2d.args[1]
 
         # check if sum of paddings from pad() and conv2d() satisfies CMSIS-NN constraints
         can_pad_be_fused = True
@@ -186,30 +182,7 @@ def check_qnn_conv2d_pad(pattern):
             pad_h_diff = int(pad_bottom - pad_top)
             can_pad_be_fused = pad_w_diff in [0, 1] and pad_h_diff in [0, 1]
 
-        # kernel zero_point should be 0
-        kernel_zp = conv2d.args[3].data.numpy()
-        kernel_zp = [kernel_zp] if kernel_zp.ndim == 0 else kernel_zp
-
-        # check if depthwise Conv2D
-        kernel_layout = conv2d.attrs.kernel_layout
-        pos_o = kernel_layout.index("O")
-        groups = conv2d.attrs.groups
-        is_depthwise = False
-        if groups == int(conv2d_input.checked_type.shape[3]) and groups == int(
-            conv2d_weight.checked_type.shape[pos_o]
-        ):
-            is_depthwise = True
-
-        ret = (
-            conv2d.attrs.out_dtype == "int32"
-            and conv2d_input.checked_type.dtype == "int8"
-            and conv2d_weight.checked_type.dtype == "int8"
-            and pattern.checked_type.dtype == "int8"
-            and bias_dtype == "int32"
-            and all([zp == 0 for zp in kernel_zp])
-            and (not is_depthwise or bias_add is not None)
-            and can_pad_be_fused
-        )
+        ret = check_qnn_conv2d(pattern) and can_pad_be_fused
         return ret
 
     def qnn_fully_connected_pattern():

src/relay/backend/contrib/cmsisnn/fuse_pads.cc

@@ -75,7 +75,8 @@ class FusePadsMutator : public MixedModeMutator {
   explicit FusePadsMutator(const IRModule& mod) : mod_(mod) {}
 
  private:
-  /*!  * \brief In order to eliminate preceding nn.pad op, pad_width of nn.pad is passed onto
+  /*!
+   * \brief In order to eliminate preceding nn.pad op, pad_width of nn.pad is passed onto
    * convolution layer to update Conv2DAttrs's padding attribute. */
   void UpdateConv2DPadding(const CallNode* conv2d_call, const CallNode* pad_call,
                            Attrs* new_attrs) {